Characterization of subcortical structures during deep brain stimulation utilizing support vector machines

Abstract

In this paper we discuss an efficient methodology for the characterization of Microelectrode Recordings (MER) obtained during deep brain stimulation surgery for Parkinson's disease using Support Vector Machines and present the results of a preliminary study. The methodology is based in two algorithms: (1) an algorithm extracts multiple computational features from the microelectrode neurophysiology, and (2) integrates them in the support vector machines algorithm for classification. It has been applied to the problem of the recognition of subcortical structures: thalamus nucleus, zona incerta, subthalamic nucleus and substantia nigra. The SVM (support vector machines) algorithm performed quite well achieving 99.4% correct classification. In conclusion, the use of a computer-based system, like the one described in this paper, is intended to avoid human subjectivity in the localization of the subcortical structures and mainly the subthalamic nucleus (STN) for neurostimulation.